• Wireless Sensor Networks [EN] : Intro 6LoWPAN Contiki OS Mote Design Project TPs Virtual Machine (Instant Contiki 2.6)

    This course is an introduction on the Internet Of Things (Web 3.0). Based on a deep study of normalized protocols over IEEE 802.15.4 (ZigBee), RPL 6lowPAN (IPV6) and IETF COAP, students will learn to develop a complete Wireless Sensor Network (WSN). Main teaching presents WSN applications, their developpements and current technologies issues since last years.

    • Introduction to the Internet Of Things and its applications
    • Internet Architecture and Protocols
    • Introduction to 6lowPAN
    • Link-Layer Technologies
    • 6lowPAN Format
    • Bootstrapping
    • Security
    • Mobility & Routing
    • Application Formats and Protocols
    • System examples
  • Artificial Intelligence [EN] : Slides TP1 TP2 TP3

    This course is an introduction to Machine Learning through three modules. First one presents Reinforcement Learning in which students should implement an algorithm to allow a robot to learn walking. Second one concerns study of Neural Network models in which students will use one on the USPS database (handwritten character recognition). Finally, project of implementing a bio-inspired robot model called ANIMAT will use both Neural Network and Reinforcement Learning to build up a totally model free agent for non-markovian domain based on a research paper.

    This course should allow students to explore other Artificial Intelligence techniques for their future projects.

  • Machine Learning [EN] : Slides TP1 TP2

    Statistical Machine Learning is process of information extraction from datasets. This information is called generalization such as our eyes are able to identify that balls and apples are spheres. There are a lot of algorithms used according to information interest:  K-mean for grouping, event prediction with Bayesian Network, etc. During this course, students learn to use Artificial Neural Network and how to configure them by results analysis.

  • Implémentation of Multilayer Perceptron with octave scripts to create Optical Character Recognition (OCR) on Handwritten texts.
  • Implementation of AdaBoost with JAVA language to classify Presidential Speeches of Chirac and Mitterrand. Adaboost uses several Simple Layer Perceptrons to extract complex correlations.

At the end of this course, student should be able to use such technics in their own project.